This study investigated the impact of 15 micronutrients on endometriosis, comprising 7 mineral micronutrients and 8 vitamins. The researchers employed the dual sample MR method to assess the causal relationships between these micronutrients and vitamins with endometriosis, as depicted in Figure 1. Stringent criteria were set to carefully select single nucleotide polymorphisms (SNPs) associated with the micronutrients and vitamins as instrumental variables to uphold the core hypothesis. During the selection of instrumental variables, SNPs linked to potential confounding factors were meticulously screened and excluded to ensure the assumption of independence was met. Moreover, to prevent instrumental variables from affecting outcomes through other pathways, a pleiotropy analysis was conducted during the MR analysis or instrumental variables showing signs of pleiotropy were removed to satisfy the exclusivity hypothesis (Supplementary Figure S1).

In our study, we prioritized the GWAS catalog database¹ and FinnGen database² for their relevance to trace elements and endometriosis, and for their high data quality. Following a comprehensive literature review (29–32), we found these databases to be highly regarded for their extensive genetic and clinical data, including long-term follow-ups, and for their data quality, robustness, and credibility. Specifically, we sourced endometriosis genetic data from FinnGen database, which covers decades of longitudinal medical information and has a depth and breadth that is rare internationally. Concurrently, we collected trace element data from GWAS, a database rich in genetic markers and phenotypes, facilitating the study of genetic variations related to various traits and diseases. Both databases are regularly updated, and researchers can access and download the latest data for free through online platforms. During data selection, we applied stringent criteria based on epidemiological and genetic research standards, considering data reliability, sample size, representativeness, and accessibility to ensure the study’s transparency and reproducibility.

The summary data from genome-wide association studies for 15 micronutrients is sourced from the GWAS catalog database. These micronutrients include copper, calcium, carotene, folic acid, iron, magnesium, potassium, zinc, as well as vitamins A, B12, B6, C, D, and E. The information regarding endometriosis is obtained from the FinnGen database, as outlined in Supplementary Table S1.

We identified significant single nucleotide polymorphisms (SNPs) (p < 5×10⁻⁶) from GWAS data for 15 micronutrients. The entire gene dataset from the European Thousand Genomes Project was utilized as a reference to assess linkage disequilibrium (LD) among SNPs. Independent SNPs (r² < 0.001, distance >10,000 kb) were chosen as instrumental variables. To ensure that the selected SNPs were exclusively linked to exposure and outcomes, the PhenoScanner database was employed to scrutinize each SNP. To support the correlation hypothesis, we evaluated the strength of instrumental variables by calculating the F-statistic (F = β²/SE²), with β as the allele effect and SE as the standard error. An F-value greater than 10 indicates a robust association between instrumental variables and mineral micronutrients as well as vitamins. In cases where SNPs related to outcomes were missing, they were replaced with highly linked SNPs. SNPs without substitutes were excluded from the analysis.

The IVW method was the primary research approach, with MR Egger, Weighted median, Simple mode, and Weighted mode analyses as supplements. IVW disregards intercept terms during regression and utilizes the reciprocal of outcome variance as weights for fitting (33, 34). Conversely, MR Egger regression considers intercept terms, displaying robustness and tolerance towards outliers and complex data factors, ensuring result accuracy and reliability. Weighted median estimation remains consistent even with up to 50% ineffective instrumental variables, addressing differences in instrumental variable accuracy effectively. Using median estimates for causal effects minimizes external interference, enhancing reliability. Simple mode adjusts for genetic variation impacts on exposure, eliminating confounders and reverse causal influences. Weighted mode enhances estimation efficiency and accuracy through genetic variation weighting, especially beneficial for large datasets, delivering precise causal effect estimates. Heterogeneity assessment via MR Egger and Inverse variance weighted analysis indicates homogeneity when p > 0.05. Multiplicity is assessed with the Egger Intercept test. Sensitivity analysis involves a leave-one-out method, sequentially removing each SNP to evaluate individual impact on merged results, ensuring overall result robustness. Funnel plots help detect bias presence. Applying these methods, we identified 15 micronutrients significantly associated with endometriosis.

By utilizing multivariate MR analysis (35), which combines MR principles with statistical methods such as multiple linear regression, we aim to determine if nutrients linked to endometriosis can each have distinct causal impacts. Additionally, we will conduct heterogeneity and pleiotropy analyses concurrently to assess the reliability and robustness of the findings.

With informed consent from endometriosis patients and healthy volunteers, we measured serum levels of vitamin B6 and calcium using high-performance liquid chromatography. Firstly, prepare a series of standard solutions containing micronutrients of different concentrations to establish a standard curve for quantitative analysis of the target substance. Then, inject the prepared serum sample into the HPLC instrument and analyze it under the specified conditions. Lastly, record and calculate the vitamin B6 and calcium levels.

Statistical analyses were conducted using RStudio (version 4.3.0) and TwoSampleMR (version 0.5.7). Causal relationships between 15 micronutrients and endometriosis were established using Odds Ratios (OR) and 95% confidence intervals (CI). All statistical tests were two-tailed with a significance level set at α = 0.05.

In this study, we acquired endometriosis genetic data from FinnGen databases, comprising 111,583 controls and 22,883 patients, along with 21,270,109 SNPs. Additionally, we sourced micronutrient genetic data from the GWAS database, including copper (n = 2,603, SNPs = 2,543,646), selenium (n = 2,603, SNPs = 2,543,617), zinc (n = 2,603, SNPs = 2,543,610), vitamin A (n = 460,351, SNPs = 9,851,867) and calcium, carotene, folate, iron, magnesium, potassium, vitamin B12, vitamin B6, vitamin C, vitamin D, and vitamin E, with a uniform sample size of 64,979 and 9,851,867 SNPs (Supplementary Table S1). We will select appropriate instrumental variables from these data based on the screening criteria of instrumental variables to explore the association between micronutrient and endometriosis. According to the instrumental variable screening criteria, the number of SNPs included in each nutrient is as follows: Copper: 6 SNPs, Selenium: 6 SNPs, Zinc: 8 SNPs, Folate: 12 SNPs, Carotene: 15 SNPs, Potassium: 14 SNPs, Iron: 11 SNPs, Calcium: 19 SNPs, Magnesium: 17 SNPs, Vitamin A: 11 SNPs, Vitamin B12: 8 SNPs, Vitamin B6: 17 SNPs, Vitamin C: 10 SNPs, Vitamin D: 13 SNPs, Vitamin E: 12 SNPs. Detailed SNP information is provided in Supplementary Table S2.

The IVW analysis in Figure 2 and Supplementary Table S3 reveal a significant causal link between calcium and endometriosis (OR = 1.4834, 95% CI: 1.0747–2.0475, p = 0.0165). This indicates that an excessive amount of calcium in women’s bodies heightens the risk of developing endometriosis. Similarly, there is a notable causal link between vitamin B6 and endometriosis (OR = 1.7060, 95% CI: 1.1796–2.4672, p = 0.0045), implying that elevated levels of vitamin B6 can increase the likelihood of endometriosis in women. On the other hand, Copper, Selenium, Zinc, Folate, Carotene, Potassium, Iron, Magnesium, Vitamin A, Vitamin B12, Vitamin C, Vitamin D, Vitamin E, and other micronutrients do not exhibit a causal relationship with endometriosis (p > 0.05). Additionally, the results from MR Egger, Weighted median, Simple mode, and Weighted mode methods align closely with those of the IVW method. Heterogeneity analysis performed using MR Egger and Inverse variance weighted (Supplementary Table S4) indicates no heterogeneity among the included SNPs (p > 0.05). The pleiotropy test of Egger Intercept test (Supplementary Table S4) reveals that horizontal pleiotropy does not introduce bias into the effect size of MR analysis (p > 0.05). Leave-one-out analysis demonstrates that no individual SNP significantly influences the effect size of MR analysis (Supplementary Figure S2). Collectively, these findings suggest the MR-derived causal relationship between micronutrients and endometriosis is reliable.

A univariate MR analysis was conducted to investigate the causal effects of endometriosis-related micronutrients, specifically Calcium and vitamin B6, on endometriosis. The results revealed a significant causal relationship between calcium, vitamin B6, and endometriosis, as shown in Figure 3 and Supplementary Table S5. The intercept term of the MR Egger regression indicated no significant directed pleiotropy between SNPs in the Calcium and Vitamin B6 datasets, with p-values exceeding 0.05 (Table 1). No significant heterogeneity was detected in genetic variations related to calcium and endometriosis (Cochran’s Q = 13.084, p = 0.787) or vitamin B6 and endometriosis (Cochran’s Q = 19.838, p = 0.228) (Table 1). Consequently, inverse variance weighted (IVW) methods under fixed and random effects were separately employed to assess the causal relationship between Calcium, vitamin B6, and endometriosis. The IVW method indicated that higher Calcium levels was associated with an increased risk of endometriosis (OR = 1.4834, 95% CI: 1.0747–2.0475, p = 0.0165), and the same was true for vitamin B6 (OR = 1.7060, 95% CI: 1.1796–2.4672, p = 0.0045). The weighted median method, compared to the supplementary method, corroborated the impact of Calcium and vitamin B6 on the risk of endometriosis, underscoring the stability of the IVW method results (Figure 3). The scatter plot showcasing the potential impact of SNPs on Calcium, Vitamin B6, and endometriosis is depicted in Supplementary Figure S3, where the slope of each method signifies the magnitude of the effect evaluated by that specific method. Both individual and combined effects of Calcium and Vitamin B6 on endometriosis are illustrated in Supplementary Figure S4. Calcium exhibits 19 SNPs, and vitamin B6 has 17 SNPs associated with an increased risk of endometriosis (Supplementary Figure S5). The sensitivity analysis outcomes using the retention method are displayed in Supplementary Figure S6, indicating that no single SNP predominantly influences the overall effect.

A multivariate MR analysis was conducted to explore the potential causal effects of micronutrients, including Vitamin D, copper, zinc, calcium, and vitamin B6, on endometriosis risk. The findings revealed that Vitamin B6 (OR = 2.397, 95% CI: 1.231–4.669, p = 0.01) independently correlated with other exposure factors, leading to an elevated risk of endometriosis (Figure 4). This indicates a significant role for vitamin B6 in endometriosis development. Moreover, the analysis found no significant heterogeneity (Cochran’s Q = 23.475, p = 0.831) or pleiotropy (p = 0.369) among the genetic variations associated with endometriosis (Table 2), supporting the credibility of the multivariate MR findings. Therefore, Vitamin B6 emerged as an independent factor associated with an increased risk of endometriosis, regardless of other exposure factors.

To test the correlation hypothesis, we initially selected SNPs associated with 15 micronutrients from GWAS data with a substantial sample size. We applied a stringent genome-wide significance threshold of p < 5 × 10⁻⁶ and verified that the F-statistics for each SNP were above 10 (Supplementary Table S2). This rigorous selection process ensures the chosen SNPs are closely related to exposure, reducing the risk of weak instrument bias. Furthermore, to examine the assumption of independence, we utilized the PhenoScanner V2 tool to assess whether the SNPs were associated with any confounding factors or risk elements related to the 15 micronutrients. Our analysis indicated that none of the selected SNPs were linked to these confounding factors. Next, we performed MVMR, confirming consistent and reliable results (Figure 4 and Table 2). Lastly, to address the exclusion hypothesis, we grouped SNPs based on specific criteria to minimize LD and enhance their independence. Neither the MR Egger regression intercept nor the IVW method detected horizontal pleiotropy (p > 0.05), highlighting the robustness of these methods against such issues (Tables 1, 2 and Supplementary Table S4). Thus, estimates of causality are unlikely to be biased by this minor level of pleiotropy.

We detected the levels of Vitamin B6 and Calcium in the serum of patients with endometriosis and healthy volunteers using high-performance liquid chromatography. Compared with healthy volunteers, we observed a significant decrease in the levels of Vitamin B6 and Calcium in the serum of patients with endometriosis (Figure 5). These results indicate a significant correlation between the levels of Vitamin B6 and Calcium in serum and the occurrence of endometriosis. Additionally, our MR analysis further confirms this finding, confirming a significant association between serum levels of Vitamin B6 and Calcium and the occurrence of endometriosis. This research result provides new strategies and directions for the prevention, early diagnosis, and clinical treatment of endometriosis.